Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Blockchain: The trust layer empowering accountable AI

Combining blockchain's data integrity applications with AI's analytical capabilities enhances the transparency, security, and accountability of AI systems.

[Images: Adobe Stock / Yan]

Profile photo of Eric Piscini

BY Eric Piscini

Blockchain is emerging as the indispensable trust layer for artificial intelligence (AI), offering transformative solutions to critical challenges. Fostering trust and transparency, distributed ledger technology (DLT) can ensure the integrity of AI systems by validating the origin and authenticity of data. This is paramount, as users cannot yet be certain of what is fake or real when working with AI-generated content. 

Data accuracy, user bias, and the promotion of misinformation are concerns DLT addresses when it comes to AI. DLT is broader than blockchain, but the idea is similar: It is a flexible, scalable, immutable way to record data across a shared network of computers. DLT does not intrinsically verify data; it makes data verifiable by storing it in a public ledger in the same way that blockchain does.

However, DLT uses different consensus and validating methods that avoid the congestion and high fees of a typical blockchain network.DLT can offer businesses certainty in discerning whether data is coming from humans or what data has been generated by AI. Blockchain is the linchpin for bringing AI toward a future of accountability, inclusivity, and innovation.

I’ve spent the last decade of my career convincing enterprises to implement DLT. Even before I joined Swirlds Labs, I was less interested in the hype side of crypto and more interested in building something that endures. Now, with so much at stake from the explosive growth of AI, it’s time for a widespread call to the secure, transparent technology DLT offers for verification. 

DATA AND USER BIAS IN AI 

AI metabolizes vast amounts of data to refine understanding and performance, drawing insights from datasets. This allows AI to recognize patterns, make predictions, and perform complex tasks. However, unconscious biases present in data can seep into AI algorithms, leading to prejudicial outcomes.

Biased AI has significant implications for businesses, especially in regulated sectors. Consequences range from substantial fines to serious reputational risks. Organizations engaging with customers must be cautious while integrating AI models.

Rigorous testing is imperative for addressing potential biases. DataRobot’s State of AI Bias report found that 54% of business leaders expressed concerns about the risk of bias in AI. An overwhelming 81% advocate for government regulations to establish clear safeguards against AI bias. 

As AI becomes integrated into our everyday lives from social media to online banking, transparency and accountability are crucial. The opacity of AI decision-making processes poses ethical dilemmas, as individuals call for the right to understand the reasoning behind AI-generated outcomes that impact their lives. Fostering trust with users and regulators is at the forefront of developing this nascent technology, and it is crucial for a responsible and beneficial AI ecosystem.

While many conversations center around the data input into AI models, there are further considerations with regard to ownership of the data and content that is created as an output of AI. At present, there is no legal basis in the U.S. for humans to claim copyright or ownership of AI-generated content (or any type of content created by non-human entities or animals).

However, as legal frameworks evolve, it’s possible that DLT could play a fundamental role in enabling creators to register ownership of the output of generative AI. For example, individuals creating images, video, or text using AI could register their content on a ledger to establish ownership of each creation to ensure that it cannot be used by external parties without permission. 

LACK OF TRUST IN AI-GENERATED CONTENT

In addition to fostering transparency and minimizing bias, DLT helps us understand what information can be trusted.

The public is now engaging with AI enthusiastically, unknowingly participating in a large-scale experiment shaping the future of AI. People are exploring what kind of content detailed prompts can create. What many don’t realize is that they are training AI to create increasingly accurate and realistic content. As time goes on, this presents individual and scaling issues. 

Beyond the issues associated with misinformation, there’s cause for concern for organizations that process and deal with large volumes of data as part of daily operations. For example, how will know-your-customer verification tools, which businesses depend upon for compliance, deal with AI as it generates more realistic human faces capable of exploiting the existing verification processes? How will businesses that require users to submit written information be able to trust that this information is coming from human beings?

Leveraging DLT’s capabilities to validate data sources and enhance AI-generated content becomes essential in these scenarios. DLT can ensure that data generated by people is hashed and logged on a verifiable ledger so organizations reliant on proving the verifiability of this information can confirm its origins. These types of verification solutions can go a long way in dealing with issues like misinformation and compliance. 

If you want to leverage DLT for your business, look for security, scalablity, sustainability, and an enterprise-grade network. Distributed ledgers lack a “middleman,” meaning the integration process is dependent on the network and organization implementing it. A network without a great UX or fixed gas fees might land you with confused clients and a hefty bill at the end of a transaction. Once you find the above-listed qualities in a DLT network, don’t let go.

FINAL THOUGHTS

The convergence of blockchain and AI is inevitable. Both technologies can assist one another in dealing with vast, rich data. Blockchain allows data to be stored and shared securely, whereas AI can analyze and generate insights from this data to create significant value.

Combining blockchain’s data integrity applications with AI’s analytical capabilities enhances the transparency, security, and accountability of AI systems. Realizing the full potential of blockchain and AI synergy requires developing new use cases, increasing collaboration, and navigating the regulatory landscape sensitively to harness their combined transformative power. Clearly, blockchain can build a better, more equitable AI landscape for all, while fostering trust across the board.


Eric Piscini is the COO and CRO at Swirlds Labs.